Title: Effects of the 2018 and 2019 floods in Kerala, India on the existing multivariate statistical models

Authors: P.G. Dileep Kumar; Narayanan Viswanath; Sobha Cyrus; Benny Mathews Abraham

Addresses: School of Engineering, Division of Civil Engineering, Cochin University of Science and Technology, Kerala, India ' Department of Mathematics, Government Engineering College Thrissur, Kerala 680009, India ' School of Engineering, Division of Civil Engineering, Cochin University of Science and Technology, Kerala, India ' Albertian Institute of Science and Technology, Kerala, India; Cochin University of Science and Technology, Kerala, India

Abstract: The state of Kerala, India, experienced severe flood events during August 2018 and 2019. The aim of this paper was to study the post-flood relevance of the multiple linear regression (MLR) and adaptive neuro-fuzzy inference system (ANFIS) models formed before floods, for Kozhikode city, Kerala, India. For this, water samples were collected from 49 different locations in the above city, in September 2019. Both the existing MLR and ANFIS models were found to be less effective on post-flood data. Hence, new MLR, structural equation (SE) and ANFIS models were formed separately for severely and less severely flood-affected samples by performing bootstrapping to address the problems caused by the small datasets. The root mean square error (RMSE) and Lorenz curve were used to analyse the performance of the models. It was observed that ANFIS models performed better than MLR models.

Keywords: flood events; statistical modelling; multiple linear regression; adaptive neuro-fuzzy inference system; structural equation modelling.

DOI: 10.1504/IJHST.2024.138795

International Journal of Hydrology Science and Technology, 2024 Vol.17 No.4, pp.371 - 394

Received: 29 Dec 2021
Accepted: 21 Dec 2022
Published online: 31 May 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article